def test_can_raise_without_storage_policy(tmpdir, filepath=DATASET_SMALL_FILEPATH):
index_fio = tmpdir.join("index.dump")
documents = load_documents(filepath)
expected_inverted_index = build_inverted_index(documents)
expected_inverted_index.dump(index_fio, storage_policy=None)
with pytest.raises(AttributeError):
InvertedIndex.load(index_fio, storage_policy=3)
def test_can_dump_and_load_inverted_index_with_array_policy_parametrized(filepath, tmpdir):
index_fio = tmpdir.join("index.dump")
documents = load_documents(filepath)
expected_inverted_index = build_inverted_index(documents)
expected_inverted_index.dump(index_fio, storage_policy=StoragePolicy)
loaded_inverted_index = InvertedIndex.load(index_fio, storage_policy=StoragePolicy)
assert expected_inverted_index == loaded_inverted_index, (
"load should return the same inverted index"
)
def arguments(small_wikipedia_inverted_index, tmpdir, filepath=DATASET_SMALL_FILEPATH):
index_fio = tmpdir.join("index.dump")
documents = load_documents(filepath)
expected_inverted_index = build_inverted_index(documents)
expected_inverted_index.dump(index_fio, storage_policy=StoragePolicy)
class Arguments:
def __init__(self):
self.dataset = DATASET_SMALL_FILEPATH
self.index = index_fio
self.output = index_fio
self.query = [["a", "b"]]
return Arguments()
def small_wikipedia_inverted_index(small_sample_wikipedia_documents):
wikipedia_inverted_index = build_inverted_index(small_sample_wikipedia_documents)
return wikipedia_inverted_index
def wikipedia_inverted_index(wikipedia_documents):
wikipedia_inverted_index = build_inverted_index(wikipedia_documents)
return wikipedia_inverted_index
def test_can_build_and_query_inverted_index(wikipedia_documents):
wikipedia_inverted_index = build_inverted_index(wikipedia_documents)
doc_ids = wikipedia_inverted_index.query(["wikipedia"])
assert isinstance(doc_ids, list), (
"inverted index query should return list"
)
def test_query_inverted_index_intersect_results(dataset_tiny_fio, query, expected_res):
documents = load_documents(dataset_tiny_fio)
tiny_inverted_index = build_inverted_index(documents)
res = tiny_inverted_index.query(query)
assert sorted(res) == sorted(expected_res)
def test_can_load_without_storage_policy(tmpdir, filepath=DATASET_SMALL_FILEPATH):
index_fio = tmpdir.join("index.dump")
documents = load_documents(filepath)
expected_inverted_index = build_inverted_index(documents)
expected_inverted_index.dump(index_fio, storage_policy=None)
InvertedIndex.load(index_fio, storage_policy=None)